System and method for battery state of health budgeting

Abstract

Aspects of the present invention relate to a control system for a vehicle, the control system comprising one or more processors collectively configured to: receive first power transfer request data indicative of a first amount of power to be transferred from a battery in the vehicle to an external electrical power system; obtain a prediction of whether a state of health measurement of the battery will be within a state of health target, in dependence on the first amount of power to be transferred; and dependent on whether the predicted state of health measurement is within the state of health target for the battery, output a signal to cause the battery to transfer the first amount of power to the external electrical power system.

Description

[0001] SYSTEM AND METHOD FOR BATTERY STATE OF HEALTH BUDGETING

[0002] TECHNICAL FIELD

[0003] Aspects of the invention relate to a vehicle, a control system for a vehicle, and a method performed by a control system for a vehicle. Particularly, but non-exclusively, the present disclosure relates to battery state of health (SOH), in particular to budgeting battery state of health in vehicle-to-everything (V2X) scenarios.

[0004] BACKGROUND

[0005] Increasingly, vehicles are being designed according to Vehicle-to-Everything (V2X) specifications. As part of this, some automotive vehicles, such as hybrid electric vehicles and plug-in electric vehicles are capable of Vehicle-to-Grid V2G transfers, enabling power transfer from vehicle to grid or home, as part of load-balancing and / or grid power management procedures. These types of vehicles can communicate with an external power source such as a power grid, and receive power transfer requests to supply power to the power grid e.g. at peak usage times. In this way, the batteries of the vehicles can be used in the manner of a power storage facility for the grid.

[0006] SUMMARY OF THE INVENTION

[0007] Aspects and embodiments of the invention provide a vehicle, a control system for a vehicle; a method performed by a control system; a computer program product, and computer readable instructions as claimed in the appended claims.

[0008] As described above, V2G is an efficient power management mechanism whereby electric vehicles can transfer power back to the electrical grid during peak energy consumption periods, and conversely, store excess power during peak energy production by the grid. V2G can thus potentially help reduce peaks and troughs in power consumption. While this is advantageous to the power grid, transferring energy to and from a battery reduces the health of the battery with each cycle, and can significantly reduce the lifetime and efficiency of a battery if performed too often. It is an object of the disclosure herein to address some of these issues and to manage V2G processes in a manner that balances the needs of the grid with performance targets for the battery over its lifetime.

[0009] According to an aspect of the present invention there is provided a control system for a vehicle. The control system comprises one or more processors collectively configured to: receive first power transfer request data indicative of a first amount of power to be transferred from a battery in the vehicle to an external electrical power system; obtain a prediction of whether a state of health measurement of the battery will be within a state of health target, in dependence on the first amount of power to be transferred; and output a signal to cause the battery to transfer the first amount of power to the external electrical power system in dependence on the predicted state of health measurement being within the state of health target for the battery.

[0010] Thus, in embodiments herein, before performing a power transfer to an external electrical power system such as the grid, a prediction is made as to whether performing the transfer will result in a state of health measurement of the battery being within a target state of health for the battery. By predicting the state of health impact of each V2G transaction before it is performed, the system can determine if the impact on state of health of that transaction is within a budgeted impact, or target. It is only if the state of heath satisfies the target that the system will authorise the transfer of power to the external electrical power system and complete the power transaction. In this way, the V2G power transfer can be carried out in a controlled manner, as power transactions will only be permitted if transferring the power from the battery in the vehicle is predicted to impact the SOH of the battery within a target limit.

[0011] In an embodiment, the one or more processors may be further collectively configured to cause the first power transfer request to be rejected if the predicted state of health measurement is not within the state of health target for the battery. In these circumstances, rejecting the first power transfer request prevents unwanted degradation of the battery, beyond the target set for said battery. This ensures the battery transfers to the grid do not impact the state of health such that it impacts the overall performance and longevity of the vehicle, while permitting transfers to the grid if these will not affect the overall target performance of said battery and vehicle.

[0012] In another embodiment, the state of health target for the battery may define a state of health curve with respect to time. The one or more processors can be collectively configured to determine the predicted state of health measurement is within the state of health target for the battery if transferring the first amount of power is predicted to lead to a state of heath above the state of health curve. Battery degradation tends to be non-linear in that the state of heath can initially degrade quite rapidly and tail off with time. Thus, by defining the target in terms of a state of health curve with respect to time, such non-linear degradation over time can be accounted for.

[0013] In a further embodiment, the state of health target for the battery sets a maximum state of health degradation value in a first time period. The one or more processors may be collectively configured to: determine the predicted state of health measurement is within the state of health target for the battery if transferring the first amount of power is predicted to lead to a state of health degradation less than the maximum state of health degradation value in the first time period. By setting a maximum SOH degradation permitted per time period (e.g. maximum SOH degradation per day, week or month) allows the battery health to degrade in a predefined, “budgeted” manner, so that power transfers can be made to the grid while also ensuring overall SOH goals for the battery can be achieved.

[0014] In some embodiments, the one or more processors may be further collectively configured to: receive second power transfer request data associated with a second power transfer request for a second amount of power to be transferred from the battery in the vehicle to the external electrical power system; obtain a second prediction of whether a state of health measurement of the battery will be within the state of health target, if the second amount of power is transferred; and cause the battery to transfer the second amount of power to the external power system and reject the first power transfer request, if: the predicted second state of health measurement is within the state of health target and a second reward associated with transferring the second amount of power is greater than a first reward associated with transferring the first amount of power. In this way, rewards can be used to prioritise different power transfer requests when more than one request is possible within the target. In some examples, the one or more processors may be collectively configured to obtain the prediction using a model, wherein the model may comprise one or more of: a learning model trained using a machine learning process; a digital twin of the vehicle; a multi-dimensional look-up table; and based on one or more mathematical equations. Machine learning (ML) models have advantages in that the models can be highly accurate and can be continuously improved as actual measured SOH following a transfer can be used as further training data. ML models can be trained on data from individual vehicles and / or vehicle models to fine tune their predictions for said individual vehicle or model respectively. This can quickly lead to highly accurate models.

[0015] In an embodiment, the model may take one or more of the following as input: a discharge profile of the battery, sensor data of the battery, a current state of health measurement for the battery, one or more historical state of health measurements for the battery, data on previous transfers of power from the battery. These parameters are all causally linked to how the battery will degrade in future transfers and can be used by the models above to predict the SOH degradation that would be incurred if the requested transform were performed. By taking more than one of the above inputs, the model can provide more accurate predictions of the battery state of health.

[0016] In another embodiment, the state of health measurement output by the model may be: a change in state of health of the battery predicted to be caused by transferring the first amount of power; or a state of health and / or state of efficiency measurement for the battery following the transfer. By outputting one of the above outputs, valuable insights can be inferred on the state of health of the battery.

[0017] In a further embodiment, the one or more processors may be further collectively configured to: obtain an actual state of health measurement of the battery, following transfer of the first amount of power to the external electrical power system, and update the model with the first power transfer request and the actual state of health measurement. Continuous training in this manner can be performed to obtain highly specialised and accurate models.

[0018] In some embodiments, the model may be stored on a cloud-based server and the one or more processors may be collectively configured to obtain the prediction of the state of health by sending a request to the cloudbased server. Storing the model on the cloud-based server increases scalability and flexibility with respect to computer power and storage, while also improving data security.

[0019] In some instances, the one or more processors may be cloud-based processors and wherein the one or more processors may be collectively configured to send a message to the vehicle in order to cause the battery to transfer the first amount of power to the external electrical power system. Cloud-based processors provide greater processing power and can be accessed from anywhere with an internet connection. In another embodiment, a vehicle comprises the battery and the above control system. By having the battery and above control system in the vehicle, the battery’s charging and discharging cycles can be managed onsite, ensuring that the battery is always in good condition.

[0020] According to an aspect of the invention, there is provided a method performed by a control system fora vehicle, the method comprising: receiving first power transfer request data indicative of a first amount of power to be transferred from a battery in the vehicle to an external electrical power system; obtaining a prediction of whether a state of health measurement of the battery will be within a state of health target, in dependence on the first amount of power to be transferred; and outputting a signal causing the battery to transfer the first amount of power to the external electrical power system in dependence on the predicted state of health measurement being within the state of health target for the battery. Predicting the battery state of health mitigates the risk of unplanned battery degradation. The power transfer can be carried out safely while staying within the limits of the degradation of the battery.

[0021] In an embodiment of this aspect, a computer program product, comprising computer readable instructions which, when the program is executed by a computer cause the computer to perform the above method.

[0022] In a further embodiment of this aspect, computer readable instructions which, when executed by one or more processors, cause the one or more processors to perform the above method.

[0023] Within the scope of this application, it is expressly intended that the various aspects, embodiments, examples, and alternatives set out in the preceding paragraphs, in the claims and / or in the following description and drawings, and in particular the individual features thereof, may be taken independently or in any combination, that is, all embodiments and / or features of any embodiment can be combined in any way and / or combination, unless such features are incompatible. The applicant reserves the right to change any originally filed claim or file any new claim accordingly, including the right to amend any originally filed claim to depend from and / or incorporate any feature of any other claim although not originally claimed in that manner.

[0024] BRIEF DESCRIPTION OF THE DRAWINGS

[0025] One or more embodiments of the invention will now be described, by way of example only, with reference to the accompanying drawings, in which:

[0026] Figure 1 is a schematic system diagram of a control system according to some embodiments of the invention;

[0027] Figure 2 shows a vehicle according to some embodiments herein;

[0028] Figure 3 shows a flow chart presenting steps of a method in a control system according to some embodiments herein;

[0029] Figure 4 shows example SOH curves according to some embodiments herein;

[0030] Figure 5 shows a method performed by a control system according to some embodiments herein; and Figure 6 shows example input and output parameters to a model according to some embodiments herein. DETAILED DESCRIPTION

[0031] As described above, power transfers to and from a battery result in degradation of battery performance, which can be captured in state of health (SOH) metrics. To manage battery performance of electric vehicle e.g. in orderto achieve high performance and long-lasting electric vehicles, it is necessary to manage power transfers from the battery, in particular in V2G scenarios.

[0032] A control system 100 for a vehicle in accordance with an embodiment of the present invention is described herein with reference to Figure 1. As shown in Figure 2, the control system 100 can be installed in a vehicle 200.

[0033] The control system 100 may be generally configured (e.g. operative) to perform any methods and functions described herein, such as method 300 or process 500 described in Figures 3 and 5, respectively. The control system 100 comprises one or more processors 120 and a memory 130. The memory 130 is configured to store instruction data (e.g. such as compiled code) representing a set of instructions 140. The one or more processors 120 may be configured to communicate with the memory 130 and to execute the set of instructions 140. The set of instructions 140, when executed by the processor 120, may cause the one or more processors to perform any of the methods herein, such as the method 300 or process 400 described below.

[0034] The one or more processors (e.g. processing circuitry or logic) 120 may be any type of processor, such as, for example, a central processing unit (CPU), a Graphics Processing Unit (GPU), a Neural Processing Unit (NPU), or any other type of processing unit. The one or more processors 120 may comprise one or more subprocessors, processing units, multi-core processors or modules that are configured to work together in a distributed manner to control the node in the manner described herein.

[0035] The memory 130 of the controller 110 can be configured to store program code or instructions 140 that can be executed by the processor 120 of the controller 110 to perform the functionality described herein. The memory 130 may be configured to store any data or information referred to herein, such as for example, requests, resources, information, data, signals, or similar that are described herein. The processor 120 may be configured to control the memory 130 to store such information.

[0036] The controller 110 comprises an input means 150 and an output means 160. The input means may be configured to receive an electrical input to the control system 100. The electrical input means may be configured, for example, to receive a power transfer request from an external power system 180 (such as the grid). The power transfer request requests a first amount of power to be transferred from a battery 170 in the vehicle to the external electrical power system 180. The input means may be configured to receive electrical input from other control systems in the vehicle, for example, from a wireless gateway module, or any other module in a vehicle. The electrical input may be received via a transceiver, or any other wired or wireless connection.

[0037] As described in more detail below, the one or more processors 120 are configured to obtain a prediction of whether a state of health measurement of the battery 170 will be within a state of health target, in dependence on the first amount of power to be transferred in the power transfer request. In other words, the prediction is an indication of whether transferring the first amount of powerto the external electrical system will result in the SOH measurement for the battery being within the SOH target for said battery.

[0038] If the predicted health measurement is within the state of health target for the battery 170, the processor outputs a control output signal 160 to the battery 170 in the vehicle 200, causing the battery 170 to transfer the amount of powerto the external power system 180. Otherwise, the power transfer request is rejected if the predicted state of health measurement is not within the state of health target for the battery 170, and the processor causes the power transfer request to be rejected.

[0039] As noted above, the control system 100 can be comprised in (e.g. form part of) a vehicle such as the vehicle 200 illustrated in Figure 2. Although the vehicle 200 depicted in Figure 2 is a passenger vehicle, it will be appreciated that the processes described herein apply to a great many different types of vehicles with a battery or batteries therein. Generally, the principles herein can be applied to electric or hybrid-electric vehicles. Examples of vehicles include but are not limited to: cars, lorries, vans, scooters, bikes, trains and buggies.

[0040] The control system 100 may be comprised in a Battery Management Control Module BMCM, or any other module suitable for performing the functionality described herein. The control system 100 may be connected to the battery 170 by a wired or wireless connection.

[0041] As described below, some parts of the methods herein can be performed in the cloud 190. The control system 100 may comprise a communication facility whereby the output means 160 may send instructions or requests to the cloud 190 and the input means 150 may receive responses and / or instructions therefrom. As an example, a prediction model can be stored on a cloud-based server and the one or more processors 120 can be cloud-based processors. The one or more processors can be configured to send and receive requests to and from the cloud-based server 190.

[0042] Figure 3 illustrates a flow chart showing steps of a method 300 performed by a control system for a vehicle. The method is a method for budgeting battery state of health. In brief, the method 300 comprises receiving, at step 310, first power transfer request data indicative of a first amount of power to be transferred from the battery 170 in the vehicle 200 to the external electrical power system 180. A prediction is obtained using a predication model, at step 320, of whether a state of health measurement of the battery 170 will be within a state of health target for the battery 170, in dependence on the first amount of power to be transferred. In step 330, dependent on whether the predicted state of health measurement is within the state of health target for the battery, the method comprises outputting a signal causing the battery to transfer the first amount of power to the external electrical power system.

[0043] The method 300 may generally be performed by the control system 100, e.g. onboard the vehicle 200. Alternatively, the method 300 may be performed by a cloud-based control system that interfaces with an onboard control system of the vehicle. Alternatively still, the method 300 may be primarily performed onboard the vehicle, with the prediction being performed by a cloud-based model. The skilled person will appreciate that these are merely examples however, and that the functionality described herein can equally be performed by different modules and / or different combinations of modules to those described herein.

[0044] In more detail, in step 310, the power request data is received by the control system 100. A power transfer request may be sent to the control system from an external power system 180. External power system 180 can be a regional or national power grid, a smaller scale power system, such as that associated with a house or group of houses, or any other external power system.

[0045] In general, the external power system 180 will make a power transfer request, for example, at peak times or at other times when the external power system is not able to meet demand. In such cases, the external power system may make a request to one or more V2G vehicles or other objects connected to the external power system, requesting transfer of power from said vehicles or other objects, to the external power system.

[0046] In step 310, power request data originating from such a power transfer request is received. The power transfer request data indicates a first amount of power (which may be defined in units of power, or energy) or energy to be transferred from a battery in the vehicle to an external electrical power system. The power transfer request data may indicate a profile of power or energy over time that is requested from the vehicle. E.g. a power or energy discharge profile that is requested from the vehicle.

[0047] In step 320, the control system obtains a prediction of whether a state of health measurement of the battery will be within a state of health target, in dependence (in other words, based on) on the first amount of power to be transferred. The prediction indicates whether, if the first amount of power is transferred from the battery, the state of health target for the battery will be met (e.g. satisfied).

[0048] State of health, as used herein, refers to any measurement that can be made on a battery that reflects its utility or performance compared to an optimal performance (e.g. compared to a new battery). As an example, SOH can be defined as the ratio of the maximum battery charge to its rated capacity (e.g. the capacity that it would have had when new). This is merely one measure of battery SOH however, and other measures may equally be used, for example, based on maximum charge or discharge rate, rate of charge leakage, maximum voltage or current available from the battery, energy efficiency, or any other measure.

[0049] The target SOH can be defined in different ways. In some embodiments, the state of health target for the battery sets a maximum SOH degradation value in a first time period. For example, the target can be defined as a SOH degradation “budget” for the first time period. The first time period can be, for example, a day, a week, a month, a year, or any other time period. As such, the target can be a maximum daily / weekly / monthly / yearly budget which represents an amount of degradation that should not be exceeded in the first time period. In such embodiments, the one or more processors are collectively configured to determine the predicted state of health measurement is within the state of health target for the battery if transferring the first amount of power is predicted to lead to a state of health degradation less than the maximum state of health degradation value in the first time period. So, if the predicted SOH degradation is less than the respective budgeted amount, then the power is transferred. However, if the degradation is predicted to be greater than the budgeted amount then it will be rejected.

[0050] In other embodiments, the SOH target may be defined with respect to a SOH curve for the battery with respect to time. The SOH curve may be based on a theoretical, or average SOH expected for a battery undergoing good performance. An example is shown in Figure 4 in which the dash-dot line 402 shows a SOH curve for a particular model of battery. An ideal or “goldilocks” line can be defined that corresponds to SOH battery measurements corresponding to an average user driving a vehicle in an average manner (e.g. at average speeds, breaking profiles, with average vehicle charging habits). In real life, the SOH profile will change according to factors such as how the user of the vehicle drives the vehicle, how regularly they charge the vehicle, and how long they charge the vehicle for in each charging cycle. As an example, a real-SOH profile may deviate from the ideal or theoretical SOH line 402, as illustrated by solid line 404.

[0051] A SOH curve may be used to set the target in step 320, for example, if the vehicle has a SOH above the SOH curve for the age of the battery, and transferring the first amount of power to the external power source is predicted to result in a SOH that is still above the SOH curve forthe age of the battery, then the power transfer request is accepted and the first amount of power is transferred. If, however, transferring the first amount of power to the external power source is predicted to result in a SOH that is below the SOH curve for the age of the battery, then the request is rejected.

[0052] In the examples above, the prediction may be obtained in various ways. For example, in step 320, the prediction may be obtained from a model. Such a model may comprise one or more of: a learning model trained using a machine learning process, a digital twin of the vehicle, a multi-dimensional look-up table, and / or one or more mathematical equations.

[0053] The models herein generally take any input parameter, or combination of input parameters that are causally related to the SOH of the battery. Example input parameters include but are not limited to: a discharge profile of the battery, sensor data of the battery (e.g. such as temperature, cooling rate, Historical Irms, estimated cell Internal Resistance, Open Circuit Voltage, OCV), a state of energy (SOE) of the battery, a state of charge (SOC) of the battery, a current state of health measurement for the battery, one or more historical state of health measurements for the battery and / or data on previous transfers of power from the battery.

[0054] The model may furthertake as input parameters relating to the first power transfer request, e.g. the first amount of power or energy that has been requested and / or any other parameters relating to the request.

[0055] As noted above, the model can be a learning model, e.g. a machine learning model. The skilled person will be familiar with machine learning and methods of training a model using a machine learning process. But in brief, a machine learning model may comprise a set of rules or (mathematical) functions that can be used to perform a task related to data input to the model. Models may be taught to perform a wide variety of tasks on input data, examples including but not limited to: determining a label for the input data, performing a transformation on the input data, making a prediction or estimation of one or more parameter values based the input data, or producing any other type of information that might be determined from the input data.

[0056] In supervised machine learning, the model learns from a set of training data comprising example inputs and corresponding ground-truth (e.g. “correct”) outputs for the respective example inputs. Generally, the training process involves learning weight values of the model so as to tune the model to reproduce the ground truth output for the input data. Different machine learning processes are used to train different types of model, for example, machine learning processes such as back-propagation and gradient-descent can be used to train neural-network models.

[0057] The model herein may generally be any type of machine learning model that can be trained to take a row of data (e.g. alpha-numeric strings) related to a power transfer request, including one or more of the example input parameters described above, as input and provide an output that can be used to determine whether a state of health measurement of the battery will be within a state of health target, if the power transfer request is fulfilled. Examples of models include but are not limited to: neural network models, linear regression models and decision tree models.

[0058] As an example, a suitable open-source neural network model is available via Scikit-learn described in the paper entitled: “Scikit-learn: Machine Learning in Python”, Pedregosa et al., JMLR 12, pp. 2825-2830, 2011. Generally, the advantages described herein can be obtained using the default parameter values and settings described in the manual.

[0059] As an example, the open-source neural network model of Scikit-learn can be trained to take the following input parameters, P:

[0060] P1. Requested Power Transfer

[0061] P2. Pack Temp

[0062] P3. Cooling Rate

[0063] P4. SOE

[0064] P5. SOC

[0065] P6. Historical Irms

[0066] P7. Est. Cell Intern. Resistance

[0067] P8. Est. OCV

[0068] In this example, the neural network can be trained (e.g. using the training processes provided in Scikit-learn) to take parameter values of the input parameters Pn above and to output a prediction of the SOH of the battery, if the battery were to transfer the amount of energy or power in the power transfer request. In this example, a training data set for the neural network can be obtained from real or simulated data. For example, the training data set can comprise training data examples, each training data example comprising a value for each of the input parameters (e.g. measured in a real-life scenario) and a ground truth value of the state of health of the battery after the transfer was made (in the real-life scenario). A training data set based on real data can thus be built up by measuring SOH of real batteries following real power transfers in a wide range of real vehicles. Some example lines of training data for this example are in Table I.

[0069] Table I

[0070] The skilled person will appreciate that the accuracy of the trained model depends on the amount of training data and the variety of training examples. For example, for the most accurate models, training data can be sampled from many vehicles with drivers with different driving styles, and battery charging habits, at a wide range of stages in the battery life of the vehicle. The skilled person will further appreciate that the training data shown in Table I is merely an example, and that a wide range of training data sets can be used, with different input and output parameters to those shown in the example in Table I.

[0071] In other examples, as noted above, a digital twin model can be used to obtain the prediction in step 320. The skilled person will be familiar with digital twins, which are virtual representations of a real object or system. Digital Twins use simulation and machine learning to simulate the object or system, and real-time data is used as boundary conditions to keep the model up-to-date. In embodiments herein, a digital twin is a virtual replica of the physical vehicle used to predict the future state and behaviour of the physical battery in the vehicle. Digital twins are described in the paper by Deng et al. entitled: “A systematic review on the current research of digital twin in automotive application” Internet of Things and Cyber-Physical Systems Volume 3, 2023, Pages 180-191.

[0072] In other examples, as noted above, a multi-dimensional look-up table may be used to obtain the prediction in step 320. A multi-dimensional look-up table stores data in a multidimensional array format, that can be used to map a set of inputs to corresponding outputs. The table may be adaptive, incorporating time-varying behaviour of the physical battery into the look-up table generation. In this way, the table can receive the input and output measurements of the battery’s behaviour to dynamically create and update the content of the underlying lookup table.

[0073] In other examples, mathematical equations such as regression models can be applied to real-data to make a state of health prediction for a set of new input parameter data.

[0074] In examples where the model is based on-board the vehicle, step 320 of “obtaining a prediction” can comprise determining the prediction directly, e.g. using the on-board model. In examples where the model is cloud- based (e.g. stored in the cloud 190) then step 320 of “obtaining a prediction” can comprise sending a message to a cloud-based server, requesting the cloud-based server determine the prediction, using the cloud-based model. The message can comprise, for example, the first power transfer request data indicative of the first amount of power to be transferred from a battery in the vehicle to an external electrical power system. The message sent will also comprise data indicating the current state of health of the battery on the basis of which the cloud-based model derives its prediction. There are various advantages to cloud-based models, such as the ability to perform continuous training of the respective model and / or a reduction in vehicle-based computing storage compared to if the model is stored on the vehicle.

[0075] The prediction can be output in a variety of ways. For example, the output of the model can be an indication of a SOH degradation (e.g. a change in state-of-health) if the first transfer is performed. In other words, a change in state of health of the battery predicted to be caused by transferring the first amount of power. In other examples, the SOH measurement is a state of health and / or state of efficiency measurement for the battery following the transfer (e.g. an absolute value of SOH after the transfer rather than a change in SOH).

[0076] In other examples, the model may be trained to take information related to both the first amount of power to be transferred and the target as input. Thus, in such embodiments, a model may output a binary output (e.g. 0 - reject the transfer; 1 make the transfer). It will be appreciated that these are merely examples and that the skilled person will be able to design a wide range of models with a wide range of different input and output parameter combinations that could be used to fulfil the functionality described herein.

[0077] A signal is output at step 330, the content of which is dependent on whether the predicted state of health measurement is within the state of health target for the battery 170. If the predicted health measurement is within the state of health target for the battery 170, the signal outputted to the battery 170 in the vehicle 200 causes the battery 170 to transfer the first amount of power to the external power system 180.

[0078] Otherwise, if the predicted state of health measurement is not within the state of health target for the battery 170, the signal output rejects the transfer of the first amount of power to the external power system 180. For example, the signal output may be sent to the external power system 180, indicating that the first amount of power will not be transferred.

[0079] In some examples, additional steps may be performed. For example, if the power transfer request is accepted and the first amount of power is transferred, following transfer of the first amount of powerto the external power system 180, the processor 120 can obtain (e.g. measure) an actual SOH measurement of the battery 170 that actually occurred due to the power transfer. The actual SOH measurement and the corresponding power transfer request can be used to update the model. For example, where the model is a learning model, such as a neural network, the power transfer request and the measured SOH measurement can be used as training data for the model (the power transfer request being an example input and the measured SOH being a ground truth output for the example input). It will be appreciated that more than one power request may be received at any given time. For example, the method 300 may further comprise receiving second power transfer request data associated with a second power transfer request for a second amount of power to be transferred from the battery in the vehicle to the external electrical power system. In such examples, the method 300 may further be used to discriminate between the first and second power transfer requests and determine which, if either, of the requests should be granted. A second request received after a first request has been received and delivered is not a second request within the present context - that is a new first request, to be considered anew. A second request is in the present context is one received after a first request has been rejected, or at least before the first request has been acceded to and delivered.

[0080] In embodiments where first and second power transfer request data is received, then a second prediction may be obtained, of whether a second state of health measurement of the battery will be within the state of health target, if the second amount of power is transferred.

[0081] The first and second power transfer requests may thus be prioritised dependent on whether they result in SOH degradation within the SOH target. Furthermore, other factors may be taken into account. For example, a reward associated with performing each transfer may be taken into account. For example, the reward may be a monetary award associated with recompense given to the battery for performing a respective transfer, a reward associated with servicing requests in particular times of need, or any other types of reward that could be made to the battery, or owner of the vehicle.

[0082] In such an example, control system 100 causes the battery to transfer the second amount of power to the external power system and reject the first power transfer request, if the predicted second state of health measurement is within the state of health target and / or a second reward associated with transferring the second amount of power is greater than a first reward associated with transferring the first amount of power. That is, if the reward associated with the second request is significantly higher than the first request, then the state of health target threshold could be reduced in dependence on the reward.

[0083] As such, a priority list can also be used to prioritise different types of transactions. The table of priority is based on a metric that measures the reward for a transaction. For example, if in the period between 13:00-13:15 there is a possibility of two 15-minute transactions, the transaction that gets the highest compensation per unit of state of health lost will be transacted.

[0084] Turning now to Figure 5, which illustrates a flow chart representing a process 500 for budgeting battery state of health according to an embodiment herein. The process 500 may be performed by a control system such as the control system, 100 described above with respect to a vehicle, such as the vehicle 200.

[0085] The process 500 comprises receiving a first power request data at step 510 which corresponds to step 310 of the method of Figure 3, as described above. The first power request data indicates a first amount of power to be transferred from the battery 170 in a vehicle 200 to the external electrical power system 180. Once the power request data has been received, the processor generates a discharge profile for the battery 170 at step 520 and receives vehicle parameters from the vehicle 200 at step 530. The discharge profile can include information on battery charge over time. When a certain power (or power profile, i.e., a power that changes over time) is being drawn from the battery and put back into the grid, the battery state of charge will change based on the output power, discharging the battery over time, generating a discharge profile. In this example, the vehicle parameters comprise: battery packet temperature, cooling rate, the state of health, the state of energy, historical current root-mean-square, estimated internal resistance, and estimated open circuit voltage. However, it will be appreciated that this is merely an example and that other parameters and / or parameter combinations can equally be used.

[0086] At step 540, (which corresponds to method step 320 of Figure 3), the discharge profile and the vehicle parameters are inputted into a model to obtain a prediction of whether the state of health measurement of the battery 170 will be within a state of health target, depending on the first amount of power to be transferred. The model can be any of the types of model described above, for example, a learning model trained using a machine learning process, a digital twin of the vehicle, a multi-dimensional look-up table and / or can be based on one or more mathematical equations.

[0087] In this embodiment, the output of the model is a change in state of health of the battery that is predicted to occur if the first amount of power is transferred and in step 550, the predicted change is compared to a budget allocated for the present time period.

[0088] In other words, in this embodiment, the state of health target for the battery sets a maximum state of health degradation permitted in a first time period. And the processors determine, at step 550, whether transferring the first amount of power is predicted to lead to a state of health degradation less than the permitted maximum state of health degradation value in the first time period. For example, in an example where the first time period is a day, the maximum state of health degradation value for each day may be 0.001 kWh. This degradation value is rolling; if for the first day in the life of the vehicle, 0 kWh in state of health state of health degradation is transacted, then on the second day in the life of the vehicle, the maximum state of health degradation value will be 0.002 kWh.

[0089] At step 560, the processor determines if transferring the first amount of power is within the health target, or the budget, available for the battery 170. For example, if on a day there is a remaining health degradation budget of 0.1 kWh, and there are 15 transactions available for the day, in different times, and all of these have a combined impact of 0.09 kWh, then they can all be performed within the allocated budget for this time period. However, if the impact on the state of health of a transaction exceeds the state of health target within the period, then an additional check is carried out to determine if there is any budget remaining from previous periods which could be allocated. For instance, if on a day there is a remaining threshold of 0.1 kWh, but there is a transaction request with an impact of 0.14kWh, then the processor checks a previous period saved in memory, for example, the previous day, to determine if there is any allocated budget remaining to transfer the power and complete the transaction. In this example, if there is a budget of at least 0.04kWH, then the transaction will be authorised at step 580. However, if there is not enough budget in the previous period, the transaction will be rejected at step 570 and the signal outputted to the battery 170 in the vehicle 200 will cause the power transfer request from the external power system to be rejected.

[0090] In some scenarios there can be a more than one request at a time, for example, second power transfer request data may be received associated with a second power transfer request for a second amount of power to be transferred from the battery 170 in the vehicle 200 to the external electrical power system 180. A second prediction is obtained of whether the state of health measurement of the battery 170 will be within the state of health target if the second amount of power is transferred. If the predicted second state of health measurement is within the state of health target and a second reward associated with transferring the second amount of power is greater than a first reward associated with transferring the first amount of power, then the processor will authorise the second transaction at step 580 (e.g. preferentially over the first transaction) and the processor will cause the battery to transfer the second amount of power to the external power system, thus completing the transaction at step 590. Steps 580 and 590 correspond to step 330 of the method of Figure 3. The control system 100 will reject the first power transfer request at step 570 and the signal outputted to the battery 170 in the vehicle 200 will cause the battery 170 to reject the transfer of the first amount of power to the external power system 180. Feasibly, even if the second prediction, like the first, is outside the available budget or exceeds the state of health target, it may be authorised if the reward for the second amount is significantly higher. Indeed, the target state of health required could be adjusted depending on the reward offered for the transfer of power. This would likely lead to an increase in the target, or reduction of the budget, for future requests.

[0091] The prediction model can be further optimised (e.g. tuned or trained) at step 600. Following the transfer of the first amount of power to the external power system 180, the processor 120 obtains an actual state of health measurement of the battery 170 and updates the model with the first power transfer request and the actual state of health measurement that was measured (e.g. the actual battery degradation measured) following the transaction. In this way, ongoing training of the model or models underlying the system can be performed, which makes the models more efficient, facilitating good V2G transactions and a healthy battery.

[0092] Figure 6 shows an example model 602 that can be used to predict SOH of a battery in a vehicle according to some embodiments herein. The model 602 may be used in steps 320 or 540 of the methods 300 and 500 described above. The model 602 can be cloud-based (e.g. hosted on a server external to a vehicle), or based in a control system of a vehicle 200. The model 602 can comprise a multi-dimensional look-up table (nD LUT), one or more equations, an optimisation model and / or a neural network. In this example, the inputs to the model are: P1 -Pack Temp, P2-Cooling Rate, P3-SOE, P4-SOC, P5-Historical Irms, P6-Est. Cell Intern. Resistance, and P7-Est. OCV and information relating to a power transfer request. The outputs are a prediction of SOH and optionally SOE, for the input power transfer request.

[0093] Turning now to other embodiments, the method 300 may be embodied in a computer program. For example, a computer program product may comprise a computer readable medium, the computer readable medium having computer readable code embodied thereon. The computer readable code can be configured such that, on execution by a suitable computer (such as the one or more processors 120 of Figure 1), the computer is caused to perform the method 300 or 500.

[0094] A computer program may take different forms, for example, source code, compiled code, executable code, or any other type of code. It will be appreciated that the source code of computer programs may be written in a wide variety of different programming languages and may take different architectural designs. For example, the functionality described herein may be split across various different sub-routines. Furthermore, the skilled person will appreciate that many different ways of splitting the functionality between the different sub-routines will be possible. The sub-routines may be stored together in one executable file to form a self-contained program. Furthermore, computer programs may call external and / or standard libraries of computer code for performing certain sub-tasks associated with the functionality described herein.

[0095] In another embodiment, there is a computer program product comprising computer readable media, having stored thereon a computer program as described above. Examples of computer readable media include, but are not limited to: ROM, such as a CD ROM, a semi-conductor ROM or a magnetic recording medium such as a hard disk.

[0096] It will be appreciated that various changes and modifications can be made to the present invention without departing from the scope of the present application.

Claims

CLAIMS1. A control system for a vehicle, the control system comprising one or more processors collectively configured to: receive first power transfer request data indicative of a first amount of power to be transferred from a battery in the vehicle to an external electrical power system; obtain a prediction of whether a state of health measurement of the battery will be within a state of health target, in dependence on the first amount of power to be transferred; and dependent on whether the predicted state of health measurement is within the state of health target for the battery, output a signal to cause the battery to transfer the first amount of power to the external electrical power system.

2. A control system as in claim 1 wherein the one or more processors are further collectively configured to: cause the first power transfer request to be rejected when the predicted state of health measurement is not within the state of health target for the battery.

3. A control system as in claim 1 or 2 wherein the state of health target for the battery defines a state of health curve with respect to time; and wherein the one or more processors are collectively configured to: determine the predicted state of health measurement is within the state of health target for the battery if transferring the first amount of power is predicted to lead to a better state of health than the state of health curve.

4. A control system as in any one of the preceding claims wherein the state of health target for the battery sets a maximum state of health degradation value in a first time period; and wherein the one or more processors are collectively configured to: determine the predicted state of health measurement is within the state of health target for the battery if transferring the first amount of power is predicted to lead to a state of health degradation less than the maximum state of health degradation value in the first time period.

5. A control system as in any one of the preceding claims wherein a reward is associated with the transfer of power from the battery in the vehicle to the external electrical power system and wherein the one or more processors are further collectively configured to: receive second power transfer request data associated with a second power transfer request for a second amount of power to be transferred from the battery in the vehicle to the external electrical power system; obtain a second prediction of whether a state of health measurement of the battery will be within the state of health target, if the second amount of power is transferred; and cause the battery to transfer the second amount of power to the external power system and reject the first power transfer request, when:the second prediction indicates that the state of health measurement is within the state of health target; and / or a second reward associated with transferring the second amount of power is greater than a first reward associated with transferring the first amount of power.

6. A control system as in any one of the preceding claims wherein the one or more processors are collectively configured to obtain the prediction using a model, wherein the model comprises one or more of: a learning model trained using a machine learning process; a digital twin of the vehicle; a multi-dimensional look-up table; and one or more mathematical equations.

7. A control system as in claim 6 wherein the model takes one or more of the following as input: a discharge profile of the battery; sensor data of the battery; a current state of health measurement for the battery one or more historical state of health measurements for the battery; and data on previous transfers of power from the battery.

8. A control system as in claim 6 or 7 wherein the state of health measurement output by the model is: a change in state of health of the battery predicted to be caused by transferring the first amount of power; or a state of health and / or state of efficiency measurement for the battery following the transfer.

9. A control system as in any one of claims 6, 7 or 8 wherein the one or more processors are further collectively configured to: obtain an actual state of health measurement of the battery, following transfer of the first amount of power to the external electrical power system, and update the model with the first power transfer request and the actual state of health measurement.

10. A control system as in any one of claims 6 to 9 wherein: the model is stored on a cloud-based server and wherein the one or more processors are collectively configured to obtain the prediction of the state of health by sending a request to the cloud-based server.11 . A control system as in any one of claims 6 to 9 further comprising a communication facility for communication between the vehicle and a cloud-based computer system, wherein: the one or more processors are cloud-based processors and wherein the one or more processors are collectively configured to send a message to the vehicle in order to cause or inhibit the battery to transfer the first amount of power to the external electrical power system.

12. A vehicle comprising the battery and the control system of any one of claims 1 to 10.

13. A method performed by a control system for a vehicle, the method comprising: receiving first power transfer request data indicative of a first amount of power to be transferred from a battery in the vehicle to an external electrical power system; obtaining a prediction of whether a state of health measurement of the battery will be within a state of health target, in dependence on the first amount of power to be transferred; and dependent on whether the predicted state of health measurement is within the state of health target for the battery, outputting a signal causing the battery to transfer the first amount of power to the external electrical power system.

14. A computer program product, comprising computer readable instructions which, when the program is executed by one or more processors cause the one or more processors to perform the method of claim 13.

15. Computer readable medium comprising instructions which, when executed by a computer, cause the computer to perform the steps of the method of claim 13.

Images